Fitting Reactive Potential Energy Surfaces with Machine Learning
Friday, June 13, 2014, 2:30 PM
I will showcase our recent work in laying down a computational framework for fitting potential energy surfaces of small assemblies of small molecules using Gaussian process regression. The interpolations are essentially automatic with very few parameters and the process robust. Examples include the water dimer and trimer, and water interacting with CO2, CH4 and NO3-. The extension to reactive potential energy surfaces will be outlined.